GitLab

* Replace einsum with ops for tracing

* Fix comment

* [ASR Pipeline] Clarify return timestamps

* fix indentation

* fix ctc check

* fix ctc error message!

* fix test

* fix other test

* add new tests

* final comment

* Add attention mask and pad token warning to many of the models

* Remove changes under examples/research_projects

These files are not maintained by HG.

* Skip the warning check during torch.fx or JIT tracing

* Switch ordering for the warning and input shape assignment

This ordering is a little cleaner for some of the cases.

* Add missing line break in one of the files

* fix

* fix

---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* Register ModelOutput subclasses as supported torch.utils._pytree nodes

Fixes #25357 where DDP with static_graph=True does not sync gradients when calling backward() over tensors contained in ModelOutput subclasses

* Add test for torch pytree ModelOutput serialization and deserialization

* Add descriptive docstring to MinNewTokensLength

It addresses https://github.com/huggingface/transformers/issues/24783



* Refine the differences between `min_length` and `min_new_tokens`

* Remove extra line

* Remove extra arguments in generate

* Add a missing space
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Run the linter

* Add clarification comments

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Add mask2former fp16 support

* Clear consistency/quality issues

* Fix consistency/quality (2)

* Add integration test for mask2former (fp16 case)

* Fix code quality

* Add integration test for maskformer (fp16 case)

* Add integration test for oneformer (fp16 case)

* Remove slow decorator from fp16 tests

* Fix lint

* Remove usage of full inference and value checks for fp16

* Temporarily comment slow for {mask, mask2, one}former

* Add fp16 support to oneformer

* Revert "Temporarily comment slow for {mask, mask2, one}former"

This reverts commit e5371edabd301cf56079def0421a0a87df307cb0.

* Remove dtype conversion noop

* Add Description And Example to Docstring

* make style corrections

* make style

* Doc Style Consistent With HF

* Apply make style

* Modify Docstring

* Edit Type in Docstring

* Feedback Incorporated

* Edit Docstring

* make style

* Post Review Changes

* Review Feedback Incorporated

* Styling

* Formatting

* make style

* pep8

* First draft

* Deal with progress bars

* Update src/transformers/utils/hub.py
Co-authored-by: Lucain <lucainp@gmail.com>

* Address review comments

* Forgot one

* Pin hf_hub

* Add argument for push all and fix tests

* Fix tests

* Address review comments

---------
Co-authored-by: Lucain <lucainp@gmail.com>

Remove Marian hack

* Loosen output shape restrictions on GPT-style models

* Use more self-explanatory variables

* Revert "Use more self-explanatory variables"

This reverts commit 5fd9ab39119558b7e750f61aa4a19014dccc5ed5.

* Generalize CFG to allow for positive prompts

* Add documentation, fix the correct class

* Remove jnp.DeviceArray since it is deprecated.

* Replace all instances of jnp.DeviceArray with jax.Array

* Update src/transformers/models/bert/modeling_flax_bert.py

---------
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* temp

* update

* update

* update

* small dim

* small dim

* small dim

* fix

* update

* fix

* fix

* fix

* fix

* fix

* fix

* fix

---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* Document check copies better and add tests

* Include header in check for copies

* Manual fixes

* Try autofix

* Fixes

* Clean tests

* Finalize doc

* Remove debug print

* More fixes

* Deal better with nested configs

* Fixes

* More fixes

* Fix last test

* Clean up existing configs

* Remove hack in MPT Config

* Update src/transformers/configuration_utils.py
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

* Fix setting a nested config via dict in the kwargs

* Adapt common test

* Add test for nested config load with dict

---------
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

* Add offline mode for agents

* Disable second check too

The former spelling is deprecated and has been discouraged for a
while. The latter spelling seems to be more common in this project
anyway, so this change ought to be safe.

Fixes https://github.com/huggingface/transformers/issues/25283

* [JAX] Bump min version

* make fixup

* Add timeout parameter to load_image function.

* Remove line.

* Reformat code
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Add parameter to docs.

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* add generate method to SpeechT5ForTextToSpeech

* update speecht5forTTS docstrings

* Remove defaults to None in generate docstrings
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update InstructBLIP values
Note: the tests are not independent. Running the test independentely produces different logits compared to running all the integration tests

* Update test values after rescale update

* Remove left over commented out code

* Revert to previous rescaling logic

* Update rescale tests

* Update list of logging integrations in docstring

Also update type hint

* Also add 'flyte' to report_to callback list

* Revert 'report_to' type hint update

Due to CLI breaking

* [MMS] Fix mms

* [MMS] Fix mms

* fix mms loading

* Apply suggestions from code review

* make style

* Update tests/models/wav2vec2/test_modeling_wav2vec2.py

Fix bug in InstructBlip generate function

Previously, the postprocessing conducted on generated sequences in InstructBlip's generate function assumed these sequences were tensors (i.e. that `return_dict_in_generate == False`).

This commit checks whether the result of the call to the wrapped language model `generate()` is a tensor, and if not attempts to postprocess the sequence attribute of the returned results object.

* added example and modified docs for EtaLogitsWarper

* make style

* fixed styling issue on 544

* removed error info and added set_seed

* Update src/transformers/generation/logits_process.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/generation/logits_process.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* updated the results

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update modeling_deformable_detr.py

Fix bugs for two stage training

* Update modeling_deformable_detr.py

* Add test_two_stage_training to DeformableDetrModelTest

---------
Co-authored-by: yupeng.jia <yupeng.jia@momenta.ai>

* resolving zero3 init when using accelerate config with Trainer

* refactor

* fix

* fix import

* add pathname and lineno to logging formatter in debug mode

* use TRANSFORMERS_VERBOSITY="detail" to print pathname and lineno

* add test for `get_keys_to_not_convert`

* add minimum patch to keep mpt lm_head from 8bit quantization

* add reivsion to

Move dtype recasting to match torchvision ToTensor

* fix detr weird issue

* Update src/transformers/models/conditional_detr/modeling_conditional_detr.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* fix copies

* fix copies

---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

[`Docs`/`quantization`] Clearer explanation on how things works under the hood. + remove outdated info (#25216)

* clearer explanation on how things works under the hood.

* Update docs/source/en/main_classes/quantization.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/main_classes/quantization.md
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* add `load_in_4bit` in `from_pretrained`

---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* fix pix2struct cross attention

* fix torchscript slow test

make build_mpt_alibi_tensor a method of MptModel so that deepspeed could override it to make autoTP work
Signed-off-by: Wang, Yi A <yi.a.wang@intel.com>

* revisit device check for pipeline

* let's raise an error.